dsRNA Dicer siRNA

Argonaute Journeys into the Heart of RISC
Erik J. Sontheimer and Richard W. Carthew


n Sophocles’ Oedipus the King, the citizens of Thebes are cursed by a plague that will end only when the murderer of Laius, their former ruler, is identified and banished. The culprit is eventually revealed to be none other than the current king (and Laius’ son), Oedipus. Although there are many things to be learned from this drama, one conclusion relates directly to the process of discovery: The object of an intensive search sometimes turns out to be hiding in plain view. So it is with the study of RNA interference (RNAi), as illustrated by two papers in this issue by Liu et al. on page 1437 (1) and by Song et al. on page 1434 (2). These authors now identify the long-sought catalytic subunit that executes RNAi, and show that it has been staring us in the face for years. In most eukaryotes, RNAi is one of several mechanisms that silence the expression of specific genes in response to double-stranded RNA (dsRNA) (3). Within cells, the dsRNA silencing triggers are cleaved into 21- to 23-nucleotide short interfering RNA (siRNA) fragments. These fragments then associate with a large protein assembly called the RNA-induced silencing complex (RISC) (see the figure). An siRNA within RISC recognizes specific messenger RNAs (mRNAs) through base pairing, and in this way guides RISC to the appropriate targets. The complex harbors a catalytic activity that specifically cleaves the bound mRNA without affecting the guide siRNA. RISC was first discovered 4 years ago (4), and the race to identify its resident proteins—especially the catalytic subunit “Slicer”—has been raging ever since. The first protein subunit discovered in RISC was Argonaute2 (5), one of a family of Argonaute proteins. Members of the Argonaute family are defined by the presence of PAZ and PIWI domains (6). The PAZ domain was recently shown to be important for binding to RNA (probably siRNA) (7), but beyond that the biochem-


The authors are in the Department of Biochemistry, Molecular Biology, and Cell Biology, Northwestern University, Evanston, IL 60208, USA. E-mail:;

ical functions of Argonaute proteins within RISC remained elusive. The Hannon and Joshua-Tor laboratories have now used a battery of biophysical, biochemical, and genetic approaches to examine Argonaute structure and function (1, 2). In doing so, they have lifted our understanding of RNAi to a new level. Most notably, they provide compelling evidence that human Argonaute2 (and, by extension, other Argonaute proteins in different species) is the elusive Slicer subunit. Song et al. (2) have solved the crystal structure of a complete Argonaute ortholog from the archaebacterium Pyrococcus furiosis. The PAZ domain sits atop a crescent composed of three other domains (including PIWI) and matches the structures of other well-characterized PAZ domains. Most tellingly, the tertiary structure of the PIWI domain clearly resembles that of ribonuclease H (RNase H) enzymes, which cleave the RNA strand of RNA/DNA hybrid duplexes. The structural similarity includes three carboxylate residues that are thought to bind and position a catalytically important divalent metal ion. The proposed catalytic site lies at the edge of a positively charged groove that extends into the PAZ domain, providing a plausible binding site for the siRNA/mRNA substrate duplex. Because RISC and RNase H are both metal-dependent enzymes that cleave one specific strand of a nucleic acid duplex and leave chemically similar termini in the products, the structural similarity immediately suggests that the PIWI domain may harbor RISC’s target mRNA cleavage activity. Of course, functional conclusions require functional data, and to this end Liu et al. (1) generated mammalian cells that express epitope-tagged versions of four human Argonaute proteins. Although siRNAs and miRNAs (microRNAs; endogenous RNAs that silence gene expression via translational control) bind to all four tagged proteins, only Argonaute2 was associated with cleavage of target mRNAs. This suggests that the other Argonautes might operate in different forms of RNA silencing. The cleavage activity remained even when the immunoprecipitates were washed under harsh conditions, indicating that it associSCIENCE VOL 305

Ago Target mRNA

mRNA degradation

An endonuclease in plain sight. The RNA interference pathway of gene silencing culminates in target mRNA cleavage by the PIWI domain of Argonaute. Within most eukaryotic cells, dsRNA molecules (red) can be cleaved by the ribonuclease Dicer (blue) into 21- to 23-nucleotide fragments called siRNAs. The siRNAs assemble into the RNA-induced silencing complex (RISC), which includes a member of the Argonaute (Ago) protein family (green). RISC assembly is accompanied by siRNA unwinding, which enables the siRNA within RISC to recognize the mRNA target (black). The PIWI domain of human Argonaute2 appears to act as an endonuclease (scissors) that cleaves the mRNA strand within the siRNA/mRNA duplex. Other nucleases then complete the mRNA degradation process.

ates tightly with Argonaute2 and may reside within Argonaute2 itself. Similar conclusions have also been reported recently by Meister et al. (8). Liu et al. further demonstrate that Argonaute2 is essential for mouse embryonic development. Mice deficient in Argonaute2 display multiple abnormalities including defects in neural tube closure and heart development. Furthermore, cells cultured from Argonaute2-deficient mice fail to mount an RNAi response upon siRNA transfection, consistent with a role for Argonaute2 in RISC.



Published by AAAS

The crystal structure of Argonaute suggests that these proteins are nucleases, and functional data indicate that mRNA target cleavage activity is associated specifically with Argonaute2. Thus, Liu et al. proceeded to map some of the determinants of cleavage activity in Argonaute2. Several point mutations within the PIWI domain specifically blocked cleavage of target mRNAs without affecting Argonaute2 protein expression or siRNA binding. Notable among these are two of the three carboxylate residues proposed to bind to a catalytic metal ion. As the authors themselves note, these results do not formally prove that those residues constitute part of an active site; the mutations could block activity by disrupting an interaction with a separate enzymatic subunit. Nonetheless, when the mutagenesis results are combined with

other functional data as well as with the structural similarity of RNase H to the PIWI domain, it makes for a compelling argument that human Argonaute2 is, in fact, the longsought Slicer subunit of RISC. Thus, the answer to one of the RNAi field’s most important questions appears to be in hand. Naturally, many questions remain. What parts (if any) do Argonaute proteins play in earlier stages of the RNAi pathway, such as dsRNA cleavage and RISC assembly? Given that specific Argonautes are essential for distinct silencing pathways—including those that affect protein synthesis (9) and heterochromatin assembly (10)— how is functional specificity established for the different Argonaute family members? Do all Argonautes require or contain nuclease activity as part of their normal du-

ties? A new entry into the nuclease lexicon is a reminder that a search for the familiar can lead to surprising discoveries.
1. J. Liu et al., Science 305, 1437 (2004); published online 29 July 2004 (10.1126/science.1102513). 2. J.-J. Song, S. K. Smith, G. J. Hannon, L. Joshua-Tor, Science 305, 1434 (2004); published online 29 July 2004 (10.1126/science.1102514). 3. G. J. Hannon, Ed., RNAi: A Guide to Gene Silencing (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, 2003). 4. S. M. Hammond, E. Bernstein, D. Beach, G. J. Hannon, Nature 404, 293 (2000). 5. S. M. Hammond, S. Boettcher, A. A. Caudy, R. Kobayashi, G. J. Hannon, Science 293, 1146 (2001). 6. M. A. Carmell, Z. Xuan, M. Q. Zhang, G. J. Hannon, Genes Dev. 16, 2733 (2002). 7. L. Joshua-Tor, Structure 12, 1120 (2004). 8. G. Meister et al., Mol. Cell 15, 185 (2004). 9. D. P. Bartel, Cell 116, 281 (2004). 10. S. I. S. Grewal, J. C. Rice, Curr. Opin. Cell Biol. 16, 230 (2004).

Crystalline Electron Pairs
Marcel Franz

t low temperatures, many compounds exhibit superconductivity, a state of vanishing electrical resistance. In high-temperature cuprate superconductors, the superconducting transition temperature Tc can be as high as 160 K. The mechanism through which this happens remains shrouded in mystery. The nature of the electronic state outside of the “superconducting dome” may hold some clues (see the figure). Much attention has been focused on the region intermediate between the antiferromagnetic Mott insulator (a state where spins are alternating) and the superconductor. Theoretical considerations predict two basic scenarios for the electron behavior in this region: Electrons either can form exotic liquid phases with “fractionalized” elementary excitations and no broken symmetries (1), or can organize into more conventional ordered states. The former possibility created a great deal of excitement among the researchers, but no convincing evidence for such exotic forms of electronic matter has been found. Recent high-resolution scanning tunneling microscopy (STM) measurements of Hanaguri et al. (2) performed on the cuprate Ca2–xNaxCuO2Cl2 (known as NaCCOC) offer an exciting glimpse of what lies to the left of the superconducting dome. They observe a periodic pattern in


the local electron density of states (LDOS) whose period is independent of the energy of the tunneling electron (see the figure, inset). This observation is strongly suggestive of underlying crystalline electronic order. It comes on the heels of earlier experimental LDOS map hints (3, 4) of such static order in another cuprate, Bi2Sr2CaCu2O8+δ (known as BiSCCO). There are essentially three ways to suppress superconductivity in a material (see arrows in the figure): Cu ion raise the temperature above LDOS maximum
Phase diagram of cuprates. AF, antiferromagnetic state; dSC, d-wave superconducting state.Arrows indicate different ways to exit from the superconducting state (2–5). (Inset) Experimental LDOS pattern of Hanaguri et al. for a single crystal with doping level x ≅ 1/8 (2).The LDOS exhibits a 4a × 4a unit cell, where a is the lattice constant. Each unit cell contains nine maxima, which (with the exception of the central one) are not registered to the Cu sites. The pattern is, however, commensurate with the underlying Cu lattice, with periodicity 4a. In contrast, the checkerboard for BiSCCO is incommensurate, with periodicity 4.3a to 4.7a (3, 4).
Magnetic field

Tc, apply a magnetic field B, or change the doping level x by altering the chemical composition. One would expect to reach the same state of electronic matter via any of the three routes, unless another phase boundary is encountered in the process. By moving along the doping axis, Hanaguri et al. (2) complete the triad of tests that convincingly demonstrate the existence of crystalline electronic order inside and outside of the superconducting dome.




Temperature [K]



T 0 0 0.1 x B 0.2

The author is in the Department of Physics and Astronomy, University of British Columbia, Vancouver, British Columbia V6T 1Z1, Canada. E-mail:

0.3 0.4




VOL 305


Published by AAAS

Sign up to vote on this title
UsefulNot useful